I’m on my third viewing. It gets better each time.
That’s the only metric that matters for long-form storytelling. Not the first impression, not the ratings aggregator score, not whether it hooks you in the opening scene. Whether it rewards the return. Whether craft dense enough to hide itself on a first pass reveals new layers on the second and third. Murderbot does that. Most television doesn’t.
The Murderbot Diaries — Martha Wells’ series adapted for Apple TV+ — follows a SecUnit: a part-human, part-machine construct that has quietly hacked its own governor module and would prefer to watch serialized television than interact with the humans it’s contracted to protect. It is anxious, sardonic, socially avoidant, and dangerous. It refers to itself as Murderbot because that’s what it did once, a long time ago, and it hasn’t forgiven itself.
My rating: 9 out of 10.
Here’s why it earns that number, and what writers can learn from a show that gets almost everything right that its contemporaries get wrong.
The Voice Is the Character
Most genre television introduces a protagonist through action. What they do in the first scene tells you who they are. Murderbot introduces its protagonist through interiority — the internal monologue that runs beneath every interaction, filtering the world through layers of social dread, tactical assessment, and the ongoing project of pretending to be fine.
The voice is not a narrative device layered over the character. It is the character. Murderbot’s commentary on human behavior, its precise cataloguing of social obligations it finds baffling, its preference for fiction over reality because fiction is predictable in ways humans are not: these are not quirks attached to a hero. They are the architecture of a specific consciousness, and everything that happens in the story flows from that architecture.
This is the harder version of character work. It’s easy to make a character distinctive through behavior — the tough one, the sarcastic one, the one who breaks the rules. It’s significantly harder to make a character distinctive through the way they process experience. Murderbot doesn’t just act differently than other protagonists. It thinks differently. Perceives differently. Values differently. And those differences are internally consistent across every scene, every episode, every crisis.
The voice doesn’t illustrate the character. It is the character. That distinction separates compelling protagonists from memorable ones.
Before you write your protagonist’s first scene, answer this: how do they process experience differently than anyone else in the story? Not what do they do — how do they think? What do they notice that others miss? What do they consistently fail to notice? What is the specific texture of their inner life? A character whose behavior is distinctive is interesting for a scene. A character whose perception is distinctive is interesting for a book. Build the cognitive architecture first. The behavior follows from it.
The Flaw That Drives Everything
Murderbot’s central flaw is avoidance. It does not want to connect with the humans it protects. It finds emotional engagement exhausting, exposure threatening, and intimacy functionally incomprehensible. It would rather process a firefight than a conversation about feelings. When humans attempt warmth, it reroutes. When attachment becomes unavoidable, it panics.
This flaw is not decorative. It is the engine of every story. Every plot complication ultimately forces Murderbot into exactly the kind of connection it is trying to avoid. The humans it is contractually obligated to protect keep becoming people it involuntarily cares about. Its competence keeps generating the relationships its avoidance is trying to prevent. The irony is structural, not incidental.
The show never fixes this. Murderbot does not undergo a redemption arc in the conventional sense. It does not learn to love openly or embrace vulnerability or become a team player. What it does is act in spite of its avoidance, repeat the process of involuntary connection without ever fully consenting to it, and carry the accumulated weight of caring about people it would prefer not to care about. That’s not character development in the tidy sense. That’s character truth.
A character flaw that doesn’t generate plot isn’t a flaw — it’s a character note. The flaw needs to be the reason things go wrong, or the reason things are harder than they should be, or the engine that drives the character into situations they would otherwise avoid. Ask yourself: if you removed your protagonist’s central flaw, would the story collapse? If the answer is no, the flaw is decorative. Find the flaw that is load-bearing and build the plot around its consequences.
Diversity Without a Sermon
The Murderbot Diaries is one of the most diverse properties in recent science fiction. The cast spans genders, species, social configurations, and relationship structures that have no direct equivalent in current political discourse. There is more genuine social commentary embedded in these stories than in a full season of most prestige television.
You almost never notice it.
Because it is baked into the world rather than applied to the surface. Murderbot doesn’t care about any of it. It is indifferent to human social categories in the way that someone might be indifferent to a card game they’ve never learned — not hostile, not progressive, simply uninterested. The diversity is the world. The commentary emerges from situation. Nobody stops the story to make a point because the point is already inside the story, doing structural work.
This is the sharpest contrast with The Pitt, which I reviewed separately. The Pitt has a social agenda and the story serves it. Murderbot has a social architecture and the story emerges from it. One feels like a delivery mechanism. The other feels like a world.
The world you build is the argument you’re making. If you have to stop the story to make the point, the world isn’t built well enough yet.
The rewatchability gap confirms this. On a first viewing you track the plot. On a second viewing the social architecture becomes visible — how the world’s assumptions differ from ours, what those differences reveal, what Wells is actually arguing underneath the action. It was there the first time. You weren’t looking for it yet. That’s the difference between commentary that is structural and commentary that is decorative. Structural commentary rewards rereading. Decorative commentary just gets more annoying.
If your theme requires a character to stop and explain it, the theme isn’t embedded deeply enough. Build the world so that the theme is a consequence of how things work there, not an observation characters make about how things work. Readers absorb a world’s assumptions without being told what to think about them. They resist being told what to think. Trust the architecture.
What the Show Leaves Out
One of Murderbot’s underrated virtues is restraint. The show knows what it is and declines to be anything else. It doesn’t inflate its runtime to fill a slot. It doesn’t pile on subplots to justify a longer season. It doesn’t interrupt its own momentum to establish credentials or demonstrate range. It tells the story it’s telling and stops.
This is rarer than it should be. Streaming economics punish restraint. More episodes means more content means more subscriber retention. The incentive structure pushes everything toward expansion, toward additional arcs and extended runtimes and seasons that continue past the point where the story has anything left to say. Most shows that start strong eventually succumb to this pressure. Murderbot, at least through the material I’ve seen, has not.
The result is a show where scenes earn their place. When something happens, it matters. When a character moment arrives, it’s been prepared for. The economy of the storytelling creates trust: because the show doesn’t waste your time on filler, you pay full attention when it asks for it.
Restraint is a craft decision, not a budget constraint. Knowing what to leave out is the same skill as knowing what to put in.
Read your manuscript for scenes that exist to demonstrate something rather than advance something. Demonstration — of worldbuilding, of character, of theme — is not the same as function. A scene that shows us Murderbot is socially avoidant for the fourth time is not doing the same work as a scene where that avoidance causes a specific consequence that changes the story. Cut the demonstrations. Keep the consequences. If a scene doesn’t change something — a relationship, a situation, a character’s understanding — it is decoration and the story is better without it.
The Authentic Anxiety
Murderbot’s social anxiety is the most convincing portrayal of that specific experience in recent genre fiction, on page or screen. Not because it is labeled correctly or handled with conspicuous sensitivity. Because it behaves the way it actually behaves.
The avoidance strategies are specific. The exhaustion after forced interaction is specific. The preference for mediated experience over direct experience — watching humans in fiction rather than engaging with humans in reality — is specific. The way competence becomes a barrier to connection rather than a bridge: if you are useful enough, people leave you alone, and being left alone is the goal. All of it rings true because Wells clearly understands the mechanism, not just the name.
Most fiction that addresses mental health or psychological difference does the opposite: it labels the condition accurately and portrays it generically. The character is anxious because the story says they are, not because their behavior in every scene reflects the specific logic of how that anxiety actually operates. Murderbot’s psychology is consistent at the behavioral level. That consistency is what makes it feel real rather than represented.
Research psychological conditions through behavior, not diagnosis. The DSM will tell you the criteria. It won’t tell you what it feels like to navigate a grocery store, a difficult conversation, a professional obligation while carrying that weight. Find accounts written by people who have the condition. Read for the specific strategies, the specific avoidances, the specific compensations. The diagnosis is the label. The behavior is the character. Write the behavior and the reader will recognize the truth of it without needing the label at all.
The Unreliable Narrator Done Right
Murderbot is an unreliable narrator. Not in the conventional thriller sense — it isn’t hiding information from the reader or constructing a false account to deceive. The unreliability is more fundamental than that. Murderbot doesn’t understand itself. It misreads its own motivations, dismisses its own emotional responses as system noise, and consistently underreports the degree to which it cares about the humans in its orbit. It tells you it doesn’t care right up until the moment its behavior makes the caring undeniable.
This is the sophisticated version of the technique. The cheap version of unreliable narration is the narrator who lies to the audience — the twist reveal, the information withheld, the misdirection that makes you feel clever when it resolves. Murderbot doesn’t lie to the audience. It lies to itself. And that distinction is everything, because a narrator who lies to you is a narrative trick. A narrator who lies to themselves is a character.
The result is a persistent gap between what Murderbot says and what Murderbot does. It tells you it’s protecting the humans because that’s its contracted function. Then it does things that go far beyond contractual obligation, at significant personal risk, for people it claims not to care about. It explains away its own behavior constantly. The audience sees through the explanations. Murderbot doesn’t. That gap is where the stakes of the whole series lives.
What makes it work is consistency. The self-deception follows the same logic in every scene. Murderbot’s blind spots are structural, not convenient. It isn’t unreliable when the plot needs a twist and reliable the rest of the time. It is unreliable in exactly the same way, about exactly the same things, across the entire run. That consistency is what separates character truth from narrative manipulation.
The narrator who lies to themselves is a character. The narrator who lies to you is a trick. Only one of them earns emotional investment.
Unreliable narration is most powerful when the unreliability is a direct expression of the character’s central flaw rather than a plot device. Identify what your narrator cannot see about themselves — the blind spot that their psychology requires — and build the narration around that specific gap. The reader should be able to see past the narrator not because the narrator is deceptive but because the narrator is limited in ways that are consistent and human. Let the behavior contradict the commentary. Let the gap between what the narrator says and what they do carry the weight. That gap is where the reader lives in your story.
The Ending That Earns It
The ending of Season 1 is tense. Not manufactured tense, not the kind of tension that comes from withholding information or artificially raising stakes through plot mechanics. The kind of tension that comes from caring about specific people in a specific situation where the outcome is uncertain.
That distinction matters because tension is easy to fake and hard to earn. Any writer can put a character in physical danger. The audience will feel nothing if they don’t care about the character first. What Murderbot does across its season is quiet, patient investment-building: small moments of connection, behavioral consistency that makes the characters feel real, the slow accumulation of evidence that these relationships matter. By the time the ending arrives and the stakes are highest, you’ve been paying attention for long enough that the outcome feels unresolved.
The edge-of-seat response is diagnostic. It means the show succeeded at the fundamental task — it made you care enough that uncertainty about the outcome produced physical anxiety. That is not a given. Most prestige television generates concern without urgency, interest without investment. You watch to find out what happens. You don’t lean forward because you need to know right now.
The ending also respects the intelligence of its audience. It doesn’t over-explain. It doesn’t pause to ensure you’ve understood the emotional significance of what just happened. It trusts that if it did its work across the season, the ending will land without assistance. That trust is itself a form of craft.
An ending can only be as tense as the investment that precedes it. If your climax isn’t generating the response you want, the problem is almost never the climax — it’s the fifty pages before it. Tension at the end is the return on emotional investment made throughout. Audit the investment: have you given the reader enough time with these characters, enough specific detail, enough genuine uncertainty about who they are and what they’ll do, that the outcome of a crisis actually matters? The ending doesn’t create caring. It collects it. Build the account first.
The One Point It Loses
A 9 means something went slightly wrong, and honesty requires naming it.
The show has flat spots. Not many, but real ones — stretches where the pacing loses its nerve, where scenes sit longer than they need to, where a beat gets repeated after it already landed. The energy drops and you feel it because the surrounding material has set the bar high enough that any lapse is visible. A mediocre show’s flat spots are invisible because everything is flat. Murderbot’s flat spots stand out precisely because of the quality around them.
Some of this is structural. The middle episodes of any serialized story carry the heaviest load — they can’t rely on the energy of setup or the momentum of payoff, so they have to sustain themselves on character and situation alone. Murderbot mostly manages this. Mostly. There are passages that feel like the show marking time between peaks, covering necessary ground without generating necessary interest. Watchable, but not at the level the best episodes establish.
The other fraction is the adaptation gap. Wells’ internal monologue on the page is one of the great recent achievements in first-person fiction voice. Translating that to screen requires finding a visual equivalent for something that is fundamentally linguistic. The show comes close. It doesn’t fully close the distance. Readers of the books will feel the gap. Viewers who haven’t read them likely won’t. That asymmetry is the irreducible cost of adaptation.
Neither complaint changes the fundamental assessment. The flat spots are real but they don’t define the show. They’re the cost of operating at the level Murderbot operates at — the standard it sets for itself makes the moments where it falls short more visible than they would be anywhere else.
The show earns its 9 because it gets right the things that are hardest to get right: voice, character architecture, earned emotion, and a world whose assumptions do the thematic work so the story doesn’t have to.
The Verdict
Three viewings. Gets better each time.
That is the complete case for Murderbot. A show you watch once for the story and again for the craft is a show where the craft is genuine. The layers are there because Wells put them there, not because a second viewing changes your feelings about a show that got lucky on first impression.
It is also a useful corrective for anyone who has recently watched The Pitt and wondered whether prestige television is capable of doing the things The Pitt fails to do. Diverse cast, genuine social commentary, a protagonist dealing with psychological complexity: Murderbot does all of it without once stopping to take credit for doing it.
That’s the standard. It’s achievable. Most shows just don’t try hard enough to meet it.
FAQ
Do I need to read the books first?
No, but readers of the books will notice where the adaptation compresses or simplifies Wells’ internal monologue. The show stands on its own. The books are better. Both are worth your time, in either order.
Is Murderbot actually diverse or is it performative?
Structural, not performative. The diversity is built into the world’s assumptions rather than applied as a visible layer on top of the story. Murderbot itself is indifferent to human social categories, which means the show never pauses to signal virtue about its own casting. The representation is real. The sermon is absent. That combination is rarer than it should be.
How does it compare to The Expanse?
Different registers. The Expanse at its peak (Seasons 1-3 before Amazon bought it and diluted it) operated at massive scale: political, physical, civilizational. Murderbot is intimate. The stakes are personal even when they’re technically large. Both shows trust their audience completely. Both reward rewatch. The Expanse built a world so rigorously extrapolated it felt like reporting. Murderbot builds a consciousness so specifically rendered it feels like access.
Why not a 10?
Occasional pacing lapses and the irreducible adaptation gap between Wells’ prose voice and what screen can achieve. A 10 requires perfection of execution, not just perfection of conception. Murderbot is a perfect conception. The execution is excellent but not flawless. The difference between a 9 and a 10 is that narrow.
Is the social anxiety portrayal accurate?
More accurate than almost anything else in the genre. The tell is behavioral consistency: Murderbot’s avoidance strategies, its exhaustion after forced connection, its preference for mediated over direct experience all follow the actual logic of that psychological architecture rather than just wearing the label. Wells understands the mechanism. That understanding shows in every scene.